Clusters data in the `CygnusObject` using specified dimensionality reduction techniques and K-means clustering. The function updates the `CygnusObject` with the resulting cluster assignments.
Usage
ClusterCygnus(
CygnusObject,
use.dims = "PCA",
relevant_markers = TRUE,
matrix_name = "scaled_exp_matrix",
clustering_method = "kmeans",
n_clusters = 3,
hdb_min_cluster_size = 10,
graph_distance = 15,
leiden_resolution = 1
)Arguments
- CygnusObject
A `CygnusObject` containing the data matrices and metadata for analysis.
- use.dims
A character string specifying the dimensionality reduction technique to use. Valid options are "PCA" (default), "tSNE", "UMAP", or None.
- relevant_markers
A logical value indicating whether to use only relevant markers for clustering (default is TRUE).
- matrix_name
A character string specifying the name of the matrix to use from the `CygnusObject`. The default is "scaled_exp_matrix".
- clustering_method
A character string specifying the method for clustering. "kmeans", "hdbscan", and "leiden" are available.
- n_clusters
An integer specifying the number of clusters to form (default is 3). Only used when clustering_method is set to "kmeans".
- hdb_min_cluster_size
An integer specifying the smallest size grouping to be considered as a cluster (default is 10). Only used when clustering_method is set to "hdmscan".
- graph_distance
The maximum number of nearest neighbours to compute. Used for leiden.
- leiden_resolution
Resolution parameter for leiden.
Details
The function performs the following steps: 1. Extracts the specified expression matrix from the `CygnusObject`. 2. Filters the matrix to include only relevant markers if specified. 3. Retrieves the selected dimensionality reduction results (`PCA`, `tSNE`, or `UMAP` or `None` ). 4. Uses the reduced dimensions for K-means clustering to assign each sample to a cluster. 5. Stores the cluster assignments in the `ev_meta` slot of the `CygnusObject`.